Jakub,
I answered your questions one by one below:
1, The heat layer's data isn't based on the number of data points. Firstly, we should set a FeatureSource (typically a shape file with point type) for a HeatLayer, there are two ways to interpret heat map's functinality/purpose.
Use column value as intensity value ( by setting HeatStyle.IntensityColumnName) .
Use predefined value as intensity value (by setting HeatStyle.PointIntensity)
The following link demonstrates how to use heat map.
wiki.thinkgeo.com/wiki/Map_S...eat_map.3F
2, The HeatMapStyle uses GDI+ to create the densities and not strict interpolation. By setting the point density the point is drawn with a circle which goes from white to black from its edge. This is done using greyscale values and after all the points are drawn then we color the greyscale using the color range specified. This is very quick but sacrifices accuracy. For spatial analysis, GRID is the appropriate tool. Although a heat map looks like a GRID, it is not one and it cannot substitute GRID for spatial analysis. If you need interpolation using IDW or something similar then I suggest you use the GridFeatureLayer I believe. This layer allows you to setup a number of data points, a grid size, an interpolation method and will generate a proper grid. You can then use another style to color the grid values.
3, 4, Currently we are still working on the spatial analysis for heat map.
Thanks,
Scott,